Answers are grounded in this paper and cite their sources.
As AI-supported mental health resources proliferate, understanding how Generative AI (GenAI) reshapes the process of well-being interventions—not just their outcomes—is critical. We conducted a 7-day randomized controlled trial comparing the Gratitude Chatbot condition (GenAI-mediated conversational journaling) against the Gratitude Journal condition (structured-prompt journaling) and an active Control condition (daily events writing). Linear mixed-effects model results revealed that, compared with the Journal condition, participants in the Chatbot condition reported less perceived effort and expressed shallower gratitude. Chatbot interactions also shifted the orientation of gratitude, eliciting more self-directed expressions. Both gratitude conditions enhanced happiness, meaningfulness, and positive linguistic expression relative to the active control, with no significant differences observed between the two gratitude formats. Thematic analysis complemented these findings, showing that participants reported feeling surprised and inspired by the GenAI-generated summaries. Our findings reveal GenAI-mediated conversational journaling as a double-edged sword: it reduces cognitive barriers, fosters inspiration, and promotes self-oriented gratitude, but may simultaneously erode reflective depth. These findings inform the design of future GenAI-supported well-being technologies that weigh the benefits and costs of AI mediation, allowing individuals to decide whether GenAI integration suits their personal needs.
No takes yet. Share an insight, caveat, or question.
Zeng et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: